Embedding New Technologies into Clinical Studies
By Exploristics Chief Data Scientific Officer – Kimberley Hacquoil
Recently, I attended the IoT Clinical Trials Europe conference on integrating IoT (the internet of things) and emerging technologies to optimise clinical trials and patient outcomes. There were many aspects covered in the one-day event, from topics around decentralised trials, simplifying the IoT eco-system, regulatory pathways and collaborations to ensure lasting adoption and evolvement of technologies. It got me thinking specifically about the opportunities for statisticians in this space; where and how can we influence and impact in these clinical trials.
Tackling the data deluge
Firstly, with so much data, it’s often hard to distinguish potential signals from all the noise. Data scientists and statisticians have efficient and automated ways to streamline the extensive data into a functional format, something that people can digest and explore to gain insights. That may be combining different data sources, summarising large datasets to create new potential endpoints for clinical trials, modelling time series data to assess which timepoints are of most interest, or summarising relationships between different aspects of the dataset. With new types of data brings opportunities for statistical method innovations as well and it’s important for statisticians to think creatively in this new space. A large amount of data does not equal an ability to solve the problem. The information needs to be abridged to deliver value to decision makers. Statisticians can help clinical teams to find clarity from the fuzziness of substantial data volumes.
Comparisons of endpoints
With new technologies and data comes new unknown and untested endpoints. It’s important to assess and validate the connection between a promising new endpoint and long-established ones which may be required from a regulatory standpoint. Statistics can quantify this relationship with the view to answering questions related to the accuracy (specificity/sensitivity) and precision, reliability, robustness, and required timepoints and timeframe for assessments. Decision-makers can be provided with vital information comparing and contrasting these different measures, which will enable more informed and efficient clinical trials. For example, how likely is a novel endpoint or diagnostic in correctly concluding a treatment effect.
Evolving clinical trial design
The use of emerging technologies can unlock potential in clinical trials with regards to accessibility, acceptability and efficiencies. We can take patient centric approaches to the next level by reducing the burden on patients whilst maintaining trial validity, reliability and consistency. There is also an opportunity to access different patient populations and broaden the reach of such trials through new technologies and remote testing.
This paradigm shift should be embraced by the clinical teams when designing trials. As new objectives and endpoints emerge, it may not be appropriate or smart to continue with the types of designs we have used before. New objectives mean reconsidering which data is really required to enable robust decision-making for the study and different decision rules should be explored. New endpoints mean re-evaluating the number of assessments and relationships between endpoints to increase the probability of a successful study. With this type of data, there may not be the necessity to have as many site visits, recruit as many patients nor open as many centres. Missing data may be something you need to scrutinise more when designing the trial to understand the impact on outcomes. With likely access to new and different patient populations, understanding differences and assessing these at the design stage will be vital for success.
All these design considerations and trade-offs can (and should) be explored simultaneously through in-silico modelling and simulation before a single patient has entered the study. This will enable smarter and more efficient study designs utilising emerging technologies.
It’s clear that utilising some technologies in clinical trials will prolong the time it takes to start a study, however, this doesn’t need to translate into an overall increased time for the clinical development program or time to market. Using a different endpoint or approach to collecting information can improve the quality of data (reduced variability, increased effect size), streamline the number of timepoints/assessments and decrease overall sample size. Technologies can also help identify patients and improve recruitment rates.
This is analogous to number of clinical trials I have been involved with, where a little more planning up front and engagement from a statistician early on in trial design ultimately leads to a better, more efficient trial with a higher probability of success. The phrase “plan well to execute fast” was used at the conference and this very much resonates with me. Statisticians have the power to navigate the clinical teams through the planning well phase to prevent rework through unclear outcomes and ambiguous trial results.
I think there are lots of exciting opportunities for data scientists and statisticians in the pharmaceutical industry to ensure these new technologies are successfully embedded into standard clinical trial practice. Technology is constantly changing and advancing, and we need to change with it to facilitate better decision-making and outcomes for our patients.